vsk marked 2 inline comments as done. vsk added inline comments.
================ Comment at: test/CodeGenCXX/ubsan-suppress-null-checks.cpp:8 + int load_member() { + // CHECK: call void @__ubsan_handle_type_mismatch + // CHECK-NOT: call void @__ubsan_handle_type_mismatch ---------------- arphaman wrote: > I think this kind of check would've passed even when this patch is not > applied. I understand that you need the first check for the single 'this' > check, but it should be rewritten to ensure that it couldn't be the check for > the `return foo`. Maybe you could use a C label before return and then check > that the 'this' check is performed before the label? Yes, this check should be tighter. This just checks that the number of checks in the method goes from 2 to 1. The label idea is great. https://reviews.llvm.org/D29530 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits